In the manufacture of semiconductor devices, various regions of a semiconductor wafer are modified by diffusing or implanting positive or negative ions (dopants), such as boron, phosphorus, arsenic, antimony and the like, into the body of the wafer to produce regions having varying conductivity. As the size of semiconductor devices becomes smaller, as in the manufacture of LSI and VLSI devices, the devices and interconnections between them are set closer together. This results in more efficient use of the wafer and increased speed of operation of the devices, but concomitantly requires more precision in the placement of the conductivity modifiers. Improvements in the equipment used to carry out the doping have also been made.
Diffusion, which involves depositing conductivity modifying ions on the surface of a wafer and driving them into the body of the wafer with heat, has limitations in establishing tight control of geometries because the diffusion process drives ions into a wafer both laterally and perpendicularly. Thus ion implantation, which can drive ions into a wafer in an anisotropic manner, has become the doping method of choice for the manufacture of modern devices.
Various ion implanters are known, using several types of ion sources. An ion beam of a preselected chemical species is generated by means of a current applied to a filament within an ion source chamber, also fitted with a power supply, ion precursor gas feeds and controls. The ions are extracted through an aperture in the ion source chamber by means of a potential between the source chamber, which is positive, and extraction means. Allied acceleration systems, a magnetic analysis stage that separates the desired ions from unwanted ions on the basis of mass and focuses the ion beam, and a post acceleration stage that ensures delivery of the required ions at the required beam current level to the target or substrate wafer to be implanted, complete the system. The size and intensity of the generated ion beam can be tailored by system design and operating conditions; for example, the current applied to the filament can be varied to regulate the intensity of the ion beam emitted from the ion source chamber. State of the art ion implantation systems have been described by Plumb et al in U.S. Pat. No. 4,754,200 and by Aitken in U.S. Pat. No. 4,578,589, both incorporated herein by reference.
The most common type of ion source used commercially is known as a Freeman source. In the Freeman source, the filament, or cathode, is a straight rod that can be made of tungsten or tungsten alloy, or other known source material such as iridium, that is passed into an arc chamber whose walls are the anode.
The arc chamber itself is fitted with an exit aperture, with means for feeding in the desired gaseous ion precursors for the desired ions; with vacuum means; with means for heating the cathode to about 2000.degree. K up to about 2800.degree. K so that it will emit electrons; with a magnet that applies a magnetic field parallel to the filament to increase the electron path length; and with a power supply connected from the filament to the arc chamber.
When power is fed to the filament, the filament temperature increases until it emits electrons that bombard the precursor gas molecules, breaking up the gas molecules so that a plasma is formed containing the electrons and various ions. The ions are emitted from the source chamber through the exit aperture and selectively passed to the target.
The filament is insulated with electrical insulators that also act to support the filament. The insulators are made of high temperature ceramic materials, such as alumina or boron nitride, that will withstand high temperatures and the corrosive atmosphere generated by precursor gas species such as BF.sub.3 or SiF.sub.4, and fragments thereof. The insulators, it turns out, severely limit the lifetime of the ion source. Although the exact number and type of ions that are generated in the source chamber are not known with certainty, various ions generated in the chamber can react both with the graphite or molybdenum walls of the chamber and with other ions in the chamber to form reaction products that deposit on the surface of the insulator, forming a conductive coating. For example, when BF.sub.3 is fed to the source chamber, chemical reactions with carbon from the graphite chamber walls and fluorine produce various carbon-fluoride species, such as CF and CF.sub.2, which further react to form a fine dust that coats the insulator. Conductive compounds may also be generated from other parts of the source chamber. Even a very thin conductive coating short circuits the arc supply and interferes with the stability of the ion beam emitted from the source chamber, eventually rendering it unusable. At this point the chamber must be cleaned and the insulators and filament reconditioned or replaced. This is the most common and most frequent cause of downtime for ion implanters.
Some prior art workers have made suggestions to prevent formation of this conductive coating on the insulators. For example, it is known to change the geometry of the electrical insulators in an arc chamber to reduce formation of the coating, but this does not greatly extend the lifetime of the unit. Others have suggested shields for the insulators to protect them from forming a conductive coating; however, the shields themselves add instabilities to the system. A cleaning discharge to etch off the coating inside the chamber has also been tried, but with mixed success since still other ions are formed during etching that can introduce other instabilities and undesired ions within the chamber.
Thus a method of reducing or eliminating the formation of a conductive coating on the filament insulators, thereby extending the time between the need for servicing the ar chamber and reducing down time for the ion implanter, would be highly desirable; further, reducing contamination of the ion beam and improving the ionization efficiency would all contribute to the economies of ion implantation.